Examples with AbstractAnalyzer org.apache.cassandra.index.sasi.analyzer.AbstractAnalyzer used on opensource projects

Example 1 with AbstractAnalyzer

use of org.apache.cassandra.index.sasi.analyzer.AbstractAnalyzer in project cassandra by apache.

the class ColumnIndex method getAnalyzer.

public AbstractAnalyzer getAnalyzer() {
    AbstractAnalyzer analyzer = mode.getAnalyzer(getValidator());
    analyzer.init(config.isPresent() ? config.get().options : Collections.emptyMap(), column.cellValueType());
    return analyzer;
}

Example 2 with AbstractAnalyzer

use of org.apache.cassandra.index.sasi.analyzer.AbstractAnalyzer in project cassandra by apache.

the class Operation method analyzeGroup.

@VisibleForTesting
protected static ListMultimap<ColumnMetadata, Expression> analyzeGroup(QueryController controller, OperationType op, List<RowFilter.Expression> expressions) {
    ListMultimap<ColumnMetadata, Expression> analyzed = ArrayListMultimap.create();
    // sort all of the expressions in the operation by name and priority of the logical operator
    // this gives us an efficient way to handle inequality and combining into ranges without extra processing
    // and converting expressions from one type to another.
    Collections.sort(expressions, (a, b) -> {
        int cmp = a.column().compareTo(b.column());
        return cmp == 0 ? -Integer.compare(getPriority(a.operator()), getPriority(b.operator())) : cmp;
    });
    for (final RowFilter.Expression e : expressions) {
        ColumnIndex columnIndex = controller.getIndex(e);
        List<Expression> perColumn = analyzed.get(e.column());
        if (columnIndex == null)
            columnIndex = new ColumnIndex(controller.getKeyValidator(), e.column(), null);
        AbstractAnalyzer analyzer = columnIndex.getAnalyzer();
        analyzer.reset(e.getIndexValue());
        // EQ/LIKE_*/NOT_EQ can have multiple expressions e.g. text = "Hello World",
        // becomes text = "Hello" OR text = "World" because "space" is always interpreted as a split point (by analyzer),
        // NOT_EQ is made an independent expression only in case of pre-existing multiple EQ expressions, or
        // if there is no EQ operations and NOT_EQ is met or a single NOT_EQ expression present,
        // in such case we know exactly that there would be no more EQ/RANGE expressions for given column
        // since NOT_EQ has the lowest priority.
        boolean isMultiExpression = false;
        switch(e.operator()) {
            case EQ:
                isMultiExpression = false;
                break;
            case LIKE_PREFIX:
            case LIKE_SUFFIX:
            case LIKE_CONTAINS:
            case LIKE_MATCHES:
                isMultiExpression = true;
                break;
            case NEQ:
                isMultiExpression = (perColumn.size() == 0 || perColumn.size() > 1 || (perColumn.size() == 1 && perColumn.get(0).getOp() == Op.NOT_EQ));
                break;
        }
        if (isMultiExpression) {
            while (analyzer.hasNext()) {
                final ByteBuffer token = analyzer.next();
                perColumn.add(new Expression(controller, columnIndex).add(e.operator(), token));
            }
        } else // "range" or not-equals operator, combines both bounds together into the single expression,
        // iff operation of the group is AND, otherwise we are forced to create separate expressions,
        // not-equals is combined with the range iff operator is AND.
        {
            Expression range;
            if (perColumn.size() == 0 || op != OperationType.AND)
                perColumn.add((range = new Expression(controller, columnIndex)));
            else
                range = Iterables.getLast(perColumn);
            while (analyzer.hasNext()) range.add(e.operator(), analyzer.next());
        }
    }
    return analyzed;
}

Example 3 with AbstractAnalyzer

use of org.apache.cassandra.index.sasi.analyzer.AbstractAnalyzer in project cassandra by apache.

the class TrieMemIndex method add.

public long add(DecoratedKey key, ByteBuffer value) {
    AbstractAnalyzer analyzer = columnIndex.getAnalyzer();
    analyzer.reset(value.duplicate());
    long size = 0;
    while (analyzer.hasNext()) {
        ByteBuffer term = analyzer.next();
        if (term.remaining() >= OnDiskIndexBuilder.MAX_TERM_SIZE) {
            logger.info("Can't add term of column {} to index for key: {}, term size {}, max allowed size {}, use analyzed = true (if not yet set) for that column.", columnIndex.getColumnName(), keyValidator.getString(key.getKey()), FBUtilities.prettyPrintMemory(term.remaining()), FBUtilities.prettyPrintMemory(OnDiskIndexBuilder.MAX_TERM_SIZE));
            continue;
        }
        size += index.add(columnIndex.getValidator().getString(term), key);
    }
    return size;
}